One of the main problems of the current Internet infrastructure is its inability to provide services at consistent quality-of-service (QoS) levels. At the same time, many emerging Internet applications, such as teleeducation, and teleconferencing, require multicast protocols that will provide the necessary QoS. In this paper, we propose QoSMIC, a multicast routing protocol for the Internet, that provides QoS-sensitive paths in a scalable, resource-efficient, and flexible way. QoSMIC differs from the previous protocols in that it identifies multiple paths and selects the one that can provide the required QoS. Two other key advantages of QoSMIC are its flexibility and adaptivity. First, the distribution tree does not have to be rooted at a preselected core router. Second, we can tradeoff between efficiency metrics depending on our needs; for example, we can tradeoff routing efficiency for a reduction in the control messages. Extensive simulations show that our protocol improves the resources utilization and the end-to-end performance compared to the current protocols. Specifically, our protocol reduces the call blocking probability by a factor of six and reduces the end-to-end delay by as much as 90% compared to the PIM protocol.

Fault recovery techniques must be reexamined in the light of the new guaranteed performance services that high-speed packet/cell switched networks will support. We investigate the rerouting of guaranteed performance service connections on the occurrence of link faults, focusing on the aspects of route selection and establishment in the network. We model the rerouting mechanism in terms of three orthogonal components: locus of reroute, reroute timing, and retry model, and explore the effect of variations in each component on the recovery process, in the presence of single- and multiple-link faults in the network. Our schemes are simulated on a cross section of network workloads, and compared using several performance criteria quantifying speed and efficacy. Our results show that global (or end-to-end) rerouting performs well as a locus of reroute selection mechanism, and that randomization is a powerful technique to determine reroute timing. We find that retrying along a new path is less effective than rerouting at a different time, and use this result to develop a new delayed retry technique. Randomization in the retry mechanism obviates the need for randomization in the initial reroute timing, so global rerouting with immediate timing and randomized retries performs well for all our performance criteria in a spectrum of network topologies and workloads for both single and multiple failures. We observe that this combination moves most of the complexity of rerouting to the edge of the network, leading to a very lightweight recovery model.

In this paper, we present, QoSMIC, a multicast protocol for the Internet that supports QoS-sensitive routing, and minimizes the importance of a priori configuration decisions (such as core selection). The protocol is resource-efficient, robust, flexible, and scalable. In addition, our protocol is provably loop-free.Our protocol starts with a resources-saving tree (Shared Tree) and individual receivers switch to a QoS-competitive tree (Source-Based Tree) when necessary. In both trees, the new destination is able to choose the most promising among several paths. An innovation is that we use dynamic routing information without relying on a link state exchange protocol to provide it. Our protocol limits the effect of pre-configuration decisions drastically, by separating the management from the data transfer functions; administrative routers are not necessarily part of the tree. This separation increases the robustness, and flexibility of the protocol. Furthermore, QoSMIC is able to adapt dynamically to the conditions of the network.The QoSMIC protocol introduces several new ideas that make it more flexible than other protocols proposed to date. In fact, many of the other protocols, (such as YAM, PIMSM, BGMP, CBT) can be seen as special cases of QoSMIC. This paper presents the motivation behind, and the design of QoSMIC, and provides both analytical and experimental results to support our claims.

The paper presents a simulation study of the use of dispersity routing to provide fault tolerance on top of a connection oriented realtime service such as that provided by the Tenet scheme. A framework to study the dispersity schemes is presented. The simulations show that the dispersity schemes, by dividing the connection's traffic among multiple paths in the network, have a beneficent effect on the capacity of the network. Thus, for certain classes of dispersity schemes, we obtain a small improvement in fault tolerance as well as an improvement in the number of connections that the network can support. For other classes of dispersity schemes, greater improvement in service may be purchased at the cost of decrease in capacity. The paper explores the tradeoffs available through exhaustive simulations. We conclude that dispersity routing is a flexible approach to increasing the fault tolerance of realtime connections, which can provide a range of improvements in service with a corresponding range of costs.

Many future applications will require guarantees on network performance, such as bounds on throughput, delay, delay jitter, and reliability. To address this need, the authors have designed, simulated, and implemented a suite of network protocols to support real-time channels (network connections with mathematically provable performance guarantees). The protocols, which constitute the prototype Tenet real-time protocol suite (Suite 1), run on a packet-switching internetwork and can coexist with the popular Internet protocol suite. The authors rely on the use of connection-oriented communication, per-channel admission control, channel rate control, and priority scheduling. This protocol suite is the first set of transport and network-layer communication protocols that can transfer real-time streams with guaranteed quality in packet-switching internetworks. The authors have performed a number of experiments and demonstrations on multiple platforms using continuous-media loads (particularly video). The results show that the approach is both feasible and practical to build, and that it can successfully provide performance guarantees to real-time applications. The paper describes the design and implementation of, the suite, the experiments performed, and some of the lessons learned.

Emerging distributed multimedia applications have stringent performance requirements in terms of bandwidth, delay, delay-jitter, and loss rate. The Tenet real-time protocol suite provides the services and mechanisms for delivering such performance guarantees, even during periods of high network load and congestion. The protocols achieve this by using resource management, connection admission control, and appropriate packet service disciplines inside the network. The Sequoia 2000 network employs the Tenet Protocol Suite at each of its hosts and routers making it one of the first wide area packet-switched networks to provide end-to-end per-connection performance guarantees. This paper presents experiments with the Tenet protocols on the Sequoia 2000 network including measurements of the performance of the protocols, the service recieved by real multimedia applications using the protocols, and comparisons with the service received by applications that use the Internet protocols (UDP/IP). We conclude that the Tenet protocols successfully protect the real-time channels from other traffic in the network, including other real-time channels, and allow channels to continue to meet their performance guarantees, even when the network is highly loaded.

The problem of finding the worst-case end-to-end delay and buffer occupancy bounds in asynchronous transfer mode (ATM) networks with rate-controlled, non-work-conserving servers is addressed. A theoretical framework is constructed to analyze such servers in isolation and in tandem. The analysis is based on a simple fluid model, but care is taken so that the computed delay and buffer occupancy values are upper bounds on actual values. A single algorithm is presented to perform these calculations in linear time. Simulation results are given in order to compare the computed worst-case delays with the actual delays obtained on some simple network topologies. The algorithm is found to predict node delays well for bursty input traffic, but poorly for smooth input traffic. Buffer requirements are predicted well in both cases.>